ai is making brad and tom fight

plus: Anthropic's $30B raise, 50% job elimination predictions, the viral AI X post, and why "the free version of ChatGPT is a flip phone."

Carley Lake

Feb 17, 2026

The real Brad and Tom. Not the AI-video fighting ones. Gareth Cattermole/Getty Images via The New York Times.

🎯 what you need to know this week in ai + marketing in 15 seconds:

"Something big is happening" - HyperWrite CEO Matt Shumer compares this AI moment to February 2020 before COVID in his viral X post. GPT-5.3 Codex and Claude Opus 4.6 now complete entire projects from plain English, test themselves, iterate - he says "I am no longer needed for the technical work of my job" (Matt Shumer)
Google tells advertisers to dump granular campaigns for AI - less manual tinkering, more feeding the system quality signals and letting it run (TechBuzz)
Anthropic raised $30B at $380B valuation - biggest venture deal of 2026. Claude Code hit $2.5B annualized revenue, business subscriptions quadrupled (Anthropic)
ByteDance’s Seedance 2.0 went viral - generates 15-second videos from text with consistent characters. Disney and Paramount sent cease-and-desist letters. A “Tom Cruise” and “Brad Pitt” AI video caused controversy (ByteDance)
Software stocks down ~$2 trillion from peak - investors worry about AI agent disruption to SaaS business models (Market news)

🗞️ top ai + marketing news

the platforms are shifting (search, ads, interfaces)

ChatGPT is citing Reddit and YouTube now. Here’s what that means.
ProFound | ~6-min read
ChatGPT is increasingly citing Reddit threads and YouTube videos in its answers. It means user-generated content isn’t just influencing people - it’s influencing the AI that influences people. My take: If your category lives on Reddit, you need to know what’s being said there. “We don’t manage that channel” is no longer a real option.

Google tells advertisers to dump granular campaigns for AI
TechBuzz | ~4-min read
Google is nudging advertisers to move away from tightly controlled, granular campaign structures and lean into AI-driven optimization instead. Less manual tinkering, more feeding the system quality signals and letting it run. My take: The edge used to be in micro-optimizing. Now it’s in creative quality and measurement clarity.

org design, governance, and what not to automate

How to design marketing organizations for AI learning and scale
MarTech | ~6-min read
Scaling AI in marketing isn’t about buying tools. It’s about building systems to test, document, and distribute learnings across teams. Structure matters more than enthusiasm. My take: If everyone is experimenting but no one is codifying, you’re not scaling. You’re just dabbling.

Before you automate marketing with AI, decide what should never be automated
Forbes | ~6-min read
A reminder to define boundaries first. Certain work - especially emotionally sensitive or reputationally risky communication - shouldn’t be handed to automation without serious oversight. My take: If you wouldn’t let an intern handle it, don’t let a model handle it alone either.

How to use AI for social media
Hootsuite | ~7-min read
A practical guide to using AI for ideation, repurposing, and analytics support in social.
My take: Social is where generic AI shows instantly. Use it to get to a draft faster, then actually add your perspective.

big money, big power moves

Software companies are reorganizing around AI
The New York Times | ~7-min read
Software firms are repositioning themselves as AI companies - or at least AI-enabled companies - as competitive pressure mounts. My take: “AI-powered” isn’t a differentiator anymore.

retail, fashion, and the trust question

Why Thrive Market’s Amina Pasha believes trust will win in an AI-first world
Modern Retail | ~5-min read
In a market saturated with synthetic content and automation, trust and transparency become core differentiators. My take: Reliability is becoming premium positioning.

Snap brings AI lenses to luxury fashion campaigns
MediaPost | ~3-min read
Luxury brands are experimenting with AI-powered AR lenses in campaigns, using tech as a layer of interactivity rather than replacing the creative core. My take: AI works when it adds capability. It struggles when it replaces taste.

Tom Cruise and Brad Pitt Appear in a Viral A.I. Fight Scene. Hollywood Is Not Amused.
The New York Times | ~6-min read
A hyper-realistic, A.I.-generated video showing digital versions of Tom Cruise and Brad Pitt in a fight scene has gone viral, prompting backlash from studios and industry groups. The clip was made using a Seedance 2.0 (see below in new tools) by Chinese technology company ByteDance. My take: The tech can do it. That doesn’t mean it should. Entertainment is about imagination, but it’s also about contracts, consent, and careers.

trends and the bigger picture

10 AI marketing trends for 2026: agentic AI and search shifts
Adweek | ~5-min read
Covers the rise of agentic workflows, changes in discovery, and how AI is reshaping the entire marketing stack. My take: The real story isn’t a single trend. It’s the fact that AI now touches every function - from planning to reporting to creative execution.

AI business startups and stunts
The Guardian | ~5-min read
A look at attention-grabbing AI startup theatrics and the fine line between innovation and spectacle. My take: Headlines are not products. At some point, you have to ship something useful.

👩‍💻 thought-leader highlights

Because it broke the internet (I haven’t written that since the Kim K. champagne Paper Magazine cover..), this week I’m just laser focused on THE X post:

Matt Shumer: “Something Big Is Happening”

@mattshumer_ | 20M+ views

HyperWrite CEO Matt Shumer compares this moment to February 2020 - when most people dismissed early COVID warnings before the world changed in three weeks. His thesis: we’re in that same “this seems overblown” phase of something much bigger.

On February 5, GPT-5.3 Codex and Claude Opus 4.6 dropped. Something clicked for him: “I am no longer needed for the actual technical work of my job.” He describes what he wants built in plain English, walks away for four hours, returns to finished work - not drafts, the actual completed thing. The AI tests its own work, iterates, and only presents it once it meets its own standards.

The key points:

AI labs focused on code first because AI building AI creates a feedback loop - each generation helps build the next, smarter version
Tech workers experienced “helpful tool” becoming “does my job better than I do” over the past year
That experience is coming to everyone else - law, finance, medicine, marketing, writing - within 1-5 years (many insiders think sooner)
OpenAI’s documentation states GPT-5.3 Codex “was instrumental in creating itself” - this isn’t a prediction, it’s already happening
Dario Amodei (Anthropic CEO) predicts 50% of entry-level white-collar jobs eliminated within 1-5 years
The free version of AI tools is over a year behind what paying users have access to - judging AI by free-tier ChatGPT is like evaluating smartphones by using a flip phone

His advice:

Pay $20/month for Claude or ChatGPT and use the best model available (not the default)
Push AI into your actual work - don’t just ask quick questions like it’s Google
The person who walks into a meeting saying, “I did this analysis in an hour instead of three days,” is the most valuable person in the room right now
Spend one hour a day experimenting with AI for six months, and you’ll understand what’s coming better than 99% of people
That window of advantage won’t stay open long

My take: I’ll be honest - the pace of change in just the past few months has surprised me.

What resonates most:

His point about the gap between public perception and reality hits home. I talk to people every week who tried ChatGPT in 2023, found it mediocre, and wrote off the whole space. Meanwhile, the people actually experimenting with current tools are producing 10x more output. That gap is real, and it’s widening fast.
The “I tried it and it wasn’t that good” crowd is working from ancient history in AI terms. Six months ago is ancient history. The models available today are genuinely different. (As you all know, I am OBSESSED with Claude Code and think everyone should use it, technical or not.)
His advice to push AI into your actual work - not just quick questions - is exactly right. The people getting ahead aren’t using it casually. They’re actively looking for ways to automate hours of work.

Where I’m more skeptical:

Do I think 50% of jobs disappear in 1-5 years? I’m not sure about that specific timeline - enterprises move slowly, compliance matters, human relationships still drive deals, and there’s a difference between capability and deployment.
The COVID comparison is emotionally effective but potentially misleading - COVID had fixed biology, AI development involves countless human decisions, and could shift direction.

But here’s what I believe completely:

The people who thrive will be the ones building AI into their workflows now
Not because AI replaces them tomorrow, but because “AI-augmented” and “non-AI-augmented” workers are becoming two very different categories with two very different levels of output
The best time to start experimenting was six months ago; the second best time is today! (Join me!!)
The window to get ahead by being early is real - and it’s closing faster than most people realize

🛠️ latest ai + marketing tools

major funding rounds

Anthropic - $30B Series G
The biggest venture deal of 2026 so far. Anthropic closed $30B at a $380B valuationn-more than double its September valuation. Claude Code’s annualized revenue hit $2.5B, and business subscriptions quadrupled since start of year.

Runway - $315M Series E
AI video startup nearly doubled valuation to $5.3B. Funding goes toward “world models”- AI systems that build internal representations of environments for planning. Follows release of Gen 4.5, which outperformed Google and OpenAI video models on benchmarks.

Simile - $100M Series A
Emerged from stealth with AI that predicts human behavior and simulates decision-making for enterprise applications—from consumer purchases to analyst questions on earnings calls.

tools that popped

Seedance 2.0 - ByteDance
ByteDance’s AI video generator went viral this week. Generates 2K video, 15-second clips from text prompts with “multi-lens storytelling” (multiple scenes, consistent characters). Disney and Paramount sent cease-and-desist letters. MPA condemned it for “unauthorized use of US copyrighted works on a massive scale.” ByteDance is now “strengthening safeguards.” Currently China-only (Jianying app), coming to CapCut globally soon.

Kuaishou Kling 3.0
Kuaishou’s competitor to Seedance - photorealistic output, 15-second videos, native audio generation across multiple languages/dialects/accents. Currently subscription-only.

other notable moves

Cowork (Anthropic) continues gaining traction - Microsoft reportedly encouraging internal teams to adopt Claude tools
Goldman Sachs rolled out Claude AI to automate accounting and compliance tasks across 12,000+ developers and back-office staff
Software stocks down ~$2 trillion from peak as investors worry about AI agent disruption to SaaS